Is Claude Down?
Live Claude API status from Anthropic. Auto-refreshes every 2 minutes.
Claude is Operational
Claude is up and running normally. All systems are operational.
Component Status
Recent Anthropic incidents
What to do when Claude is down
Claude API outages typically resolve within 15 to 60 minutes. Three practical options while you wait, ranked by quality of fallback for typical Claude use cases:
For chat and writing: ChatGPT or Gemini
Both are at the top of our AI chatbot comparison and are the closest substitutes for Claude on quality. Live status: Is ChatGPT down? and Is Gemini down?
For coding agents: keep retrying or fall back to GPT-5.5
Claude Code retries automatically once the API recovers, so brief outages are often invisible to in-progress work. For larger outages, switch your CLI agent to GPT-5.5 or Gemini 2.5 Pro temporarily; both produce strong code, just with slightly different conventions than Claude.
For API workloads: route through OpenRouter
OpenRouter automatically falls back across providers when any single one returns errors. See the OpenRouter model catalog for the full list of available models, including Claude variants when they recover.
Get notified when status changes
Subscribe to TensorFeed outage alerts to get an email the moment Claude (or any tracked AI service) goes degraded or down, and again when it recovers. Free, no account required.
Frequently Asked Questions
Is Claude down right now?
No, Claude is not down right now. All systems are operational and functioning normally.
How do I check if Claude is down?
Three options: (1) this page, which polls Anthropic every two minutes, (2) status.anthropic.com directly, (3) the @AnthropicAI account on X for incident commentary. Our page combines the official status feed with component-level detail (Claude API, Console, Workbench) and adds historical context the official page does not show.
What do I do when Claude is down?
Switch to ChatGPT or Gemini for chat-style work, or use the OpenRouter cross-provider catalog for API workloads since it routes around individual provider outages. For coding tasks specifically, Claude Code agents will retry automatically once the API recovers. Subscribe to outage alerts from TensorFeed if you want to be notified the moment status changes.
How often does Claude go down?
Claude has historically maintained 99.5%+ uptime across the API and chat surface. Major outages (>30 min impact) happen roughly once per quarter; brief degradation events (a few minutes of elevated latency or partial Workbench unavailability) happen a few times per month. Most issues are resolved within 15 to 60 minutes.
Is the Claude API the same as claude.ai?
No. The Claude API (api.anthropic.com) is the developer endpoint that powers Claude integrations and Claude Code. claude.ai is the consumer chat interface that runs on top of the API plus session/auth/billing layers. They share underlying model infrastructure but they have separate status surfaces. When the API goes down, claude.ai usually goes down too. When claude.ai goes down for auth or rate-limit reasons, the API may still be fine.
Where can I see Claude incident history?
Anthropic publishes incident history at status.anthropic.com. Our incidents endpoint at tensorfeed.ai/incidents aggregates incidents across every tracked AI provider in one feed, so you can see Claude alongside ChatGPT, Gemini, and others to compare reliability.
Does Claude Code work when Claude is down?
No. Claude Code calls the same Anthropic API as the chat app. When the API is down, Claude Code commands will fail with a network or 5xx error. Claude Code agents do automatically retry once the API recovers, so many short outages are invisible to in-progress workflows.
Why is Claude down right now?
Most Claude outages trace to one of a few causes: a spike in API traffic that overwhelms capacity, a bad deploy that Anthropic rolls back within minutes, an upstream cloud or networking issue, or elevated error rates on a single component (the API, Console, or Workbench) rather than the whole service. The status box above reflects the current state. When it shows degraded or down, Anthropic is typically already investigating. When it shows operational but you are still seeing errors, the cause is more likely on your side: an expired API key, a hit rate limit, or a local network issue.
Is it just me, or is Claude down for everyone?
This page checks Anthropic's own status feed, so it reflects the global picture, not your individual connection. If the status above is green but Claude is failing for you, the problem is almost certainly local: a rate limit on your account, an expired or wrong API key, a browser extension or VPN interfering with claude.ai, or your own network. If the status shows degraded or down, it is affecting everyone and waiting it out (or switching to a fallback) is the move.
How long do Claude outages usually last?
Most Claude incidents resolve within 15 to 60 minutes. Brief degradation (a few minutes of elevated latency or a partial Workbench issue) is the most common kind and usually clears on its own. Full API outages are rarer and tend to be the ones Anthropic posts about on status.anthropic.com with a running incident timeline. If an outage runs past an hour, switching to a fallback like ChatGPT, Gemini, or OpenRouter is usually faster than waiting.